Google launches a Gemini workplace agent that can write code and run tasks

Google has launched a Gemini-based agent aimed at workplace use, according to CBS News. The company says the agent can write code and run tasks.

Unlike assistants that mainly answer questions or autocomplete text, the positioning here emphasizes execution. The agent is meant not just to generate content but to take actions inside a task flow and complete a series of steps on the user's behalf.

The inclusion of code-writing signals that Google is bringing developer capability into the workplace agent as well. Combining code generation with task execution points toward a general-purpose work agent that can move across tools and steps, rather than an assistant tied to a single application.

The launch continues Google's push to embed Gemini in enterprise workflows. Putting agents inside productivity and development tools is becoming a key battleground as major vendors compete for enterprise customers; whoever embeds their capabilities in the software people already use has the better chance of locking users in.

The report is based on the company's own account, and the public information centers on capability claims. Concrete details about the product's form, availability, pricing, and rollout timing have not appeared in the candidate material, so how far this has actually landed remains unclear.

For enterprises, the biggest hurdle for a workplace agent is usually not whether it can write code, but permissions, security, and controllability. An agent that acts autonomously has to have clearly defined limits on what data it can access, what operations it can trigger, and how mistakes get rolled back.

What to watch next is how Google will deliver this agent to businesses, and whether writing code and running tasks holds up reliably in real workflows. Those two points will determine whether this is a feature announcement or a substantive shift in how office work gets done.